Method and apparatus for recognizing imaged information-bearing medium, computer device and medium

Abstract

A method and apparatus for recognizing an imaged information-bearing medium, a computer-readable storage device and a computer device are provided. The method comprising: acquiring a first image of the imaged information-bearing medium; performing text recognition on the first image to acquire a text content of the imaged information-bearing medium; classifying the imaged information-bearing medium to acquire a type of the imaged information-bearing medium; and archiving the text content according to the type.

What is claimed is: 
     
       1. A method for recognizing an imaged information-bearing medium, including:
 acquiring a first image of the imaged information-bearing medium comprising:
 performing target detection and correction on the imaged information-bearing medium in an original image based on the acquired original image to acquire the first image; 
 
 performing text recognition on the first image to acquire a text content of the imaged information-bearing medium; 
 classifying the imaged information-bearing medium to acquire a type of the imaged information-bearing medium; and 
 archiving the text content according to the type, 
 wherein performing text recognition on the first image to acquire a text content of the imaged information-bearing medium includes:
 performing text detection on the first image to acquire multiple second images; and 
 recognizing the multiple second images with a preset text recognition network model to acquire the text content of the imaged information-bearing medium; 
 
 wherein performing target detection and correction on the imaged information-bearing medium in an original image based on the acquired original image to acquire the first image comprises: 
 performing image binarization based on the acquired original image; 
 performing edge detection to acquire an outline of the largest rectangle in the original image, or performing straight line detection to acquire groups of a horizontal straight line set and a vertical straight line set, and merging approximate parallel lines to determine an optimal boundary and vertices of the imaged information-bearing medium; 
 segmenting the first image from the original image by perspective transformation. 
 
     
     
       2. The method of  claim 1 , wherein acquiring a first image of the imaged information-bearing medium further comprises:
 determining whether a deformation of the first image is within a preset degree of deformation, and 
 when the deformation of the first image goes beyond the preset degree of deformation, performing text direction detection on the first image and rotating the first image to correct the first image. 
 
     
     
       3. The method of  claim 2 , wherein performing text direction detection on the first image and rotating the first image to correct the first image comprises:
 performing text direction detection on the first image with a preset full-angle text detection classification model and output a rotation angle; and 
 rotating the first image by the rotation angle. 
 
     
     
       4. The method of  claim 1 , wherein performing text detection on the first image to acquire multiple second images comprises:
 acquiring axial symmetric bounding boxes according to the first image; 
 performing text score prediction on each axial symmetric bounding box to acquire area box with minimum inclination; and 
 detecting weak and small texts in the area box with minimum inclination and acquire the multiple second images. 
 
     
     
       5. The method of  claim 1 , wherein the preset text recognition network model comprises a CRNN text recognition network model, and the CRNN text recognition network model combines both a CNN convolutional neural network and an RNN recurrent neural network,
 wherein recognizing the multiple second images with a preset text recognition network model to acquire the text content of the imaged information-bearing medium includes:
 inputting the multiple second images into the CRNN text recognition network model; and 
 recognizing the multiple second images as character strings according to a process of the CNN convolutional neural network-LSTM long short-term memory network-CTC connectionist temporal classification. 
 
 
     
     
       6. The method of  claim 1 , wherein archiving the text content according to the type comprises:
 archiving the text content using a preset corresponding text archiving template according to the type, and acquiring archiving information of the imaged information-bearing medium. 
 
     
     
       7. The method of  claim 1 , wherein after acquiring the first image of the imaged information-bearing medium, and before performing text recognition on the first image to acquire a text content of the imaged information-bearing medium, the method further includes:
 performing barcode detection on the first image. 
 
     
     
       8. The method of  claim 7 , wherein performing barcode detection on the first image comprises:
 performing image rotation correction on the first image; 
 performing noise reduction processing on the corrected first image; 
 performing edge detection on the noise-reduced first image to acquire an edge image; and 
 detecting and filtering the edge image with probability transformation to determine whether there is a barcode, if not, exiting the barcode detection, and if yes, performing horizontal expansion on the line segment at the edge of the barcode to acquire a connected domain, determining a barcode area of the barcode according to the connected domain, and performing decoding processing to recognize the imaged information-bearing medium. 
 
     
     
       9. The method of  claim 1 , wherein classifying the imaged information-bearing medium to acquire a type of the imaged information-bearing medium comprises:
 performing rough classification on the imaged information-bearing medium according to an aspect ratio of the first image; and 
 performing fine classification on the roughly classified imaged information-bearing medium using a preset corresponding image classifier and based on the first image, to acquire the type of the imaged information-bearing medium. 
 
     
     
       10. The method of  claim 1 , wherein classifying the imaged information-bearing medium to acquire a type of the imaged information-bearing medium comprises:
 performing rough classification on the imaged information-bearing medium according to an aspect ratio of the first image; and 
 performing fine classification on the roughly classified imaged information-bearing medium using a preset corresponding text classifier and based on the text content, to acquire the type of the imaged information-bearing medium. 
 
     
     
       11. A computer-readable storage medium on which a computer program is stored, when the program is executed by a processor, the processor is configured to perform the method of  claim 1 . 
     
     
       12. A computer device, comprising a storage, a processor, and a computer program stored in the storage and capable of running on the processor, and when the processor executes the program, the processor is configured to perform the method of  claim 1 . 
     
     
       13. An apparatus for recognizing an imaged information-bearing medium, comprising:
 a distortion correction module configured to acquire a first image of the imaged information-bearing medium, wherein in order to acquire a first image of the imaged information-bearing medium, the distortion correction module is configured to:
 perform target detection and correction on the imaged information-bearing medium in an original image based on the acquired original image to acquire the first image; 
 
 a text recognition module configured to perform text recognition on the first image to acquire a text content of the imaged information-bearing medium; 
 an imaged information-bearing medium classification module configured to classify the imaged information-bearing medium to acquire a type of the imaged information-bearing medium; and 
 a text archiving module configured to archive the text content according to the type, 
 wherein the text recognition module is further configured to:
 perform text detection on the first image to acquire multiple second images; and 
 recognize the multiple second images with a preset text recognition network model to acquire the text content of the imaged information-bearing medium; 
 
 wherein in order to perform target detection and correction on the imaged information-bearing medium in an original image based on the acquired original image to acquire the first image, the distortion correction module is further configured to: 
 perform image binarization based on the acquired original image; 
 perform edge detection to acquire an outline of the largest rectangle in the original image, or perform straight line detection to acquire groups of a horizontal straight line set and a vertical straight line set, and merge approximate parallel lines to determine an optimal boundary and vertices of the imaged information-bearing medium; 
 segment the first image from the original image by perspective transformation. 
 
     
     
       14. The apparatus for recognizing an imaged information-bearing medium of  claim 13 , further comprising:
 a barcode detection module configured to perform barcode detection on the first image.